文章摘要
基于卷积神经网络的天然橡胶混炼胶门尼粘度的在线预测
Online Prediction of Mooney Viscosity of NR Compound based on Convolutional Neural Network
投稿时间：2024-01-30  修订日期：2024-01-30
DOI：10.12136/j.issn.1000-890X.2025.08.0615
中文关键词: NR混炼胶  卷积神经网络  门尼粘度  预测模型
英文关键词: 
基金项目:国家自然科学基金面上项目（52173101）；山东省自然科学基金重点项目（ZR2020KE037）
作者单位E-mail
李朝阳 青岛科技大学 lishuxiadezhaoyang@163.com 
边慧光 青岛科技大学  
陈海龙 山东丰源轮胎制造股份有限公司  
成　坤 山东丰源轮胎制造股份有限公司  
汪传生* 山东省高分子材料先进制造技术重点实验室；轮胎先进装备与关键材料国家工程研究中心 wcsmta@qust.edu.cn 
摘要点击次数: 200
全文下载次数: 114
中文摘要:
      为解决天然橡胶（NR）混炼胶门尼粘度[ML（1＋4）100 ℃]（简称门尼粘度）检测存在时效滞后和操作繁琐等问题，提出一种基于卷积神经网络（CNN）的NR混炼胶门尼粘度的在线预测方法。将混炼过程中密炼机记录的密炼室温度、混炼时间、物料温度、转子转速和扭矩、塑炼胶门尼粘度等参数构成矩阵并作为输入，将混炼胶门尼粘度作为输出，构建了从输入到输出的复杂映射关系的NR混炼胶门尼粘度的CNN预测模型。结果表明：NR混炼胶门尼粘度的CNN预测模型基本收敛，门尼粘度的预测值与实测值的均方误差为0.5，CNN预测模型的相关因数（ R2）为0.874 3，门尼粘度的预测误差在±根号5以内的识别率为88.46%，预测误差在±根号2以内的识别率为42.30%；CNN预测模型对塑炼胶门尼粘度为45～75范围内的NR混炼胶门尼粘度的预测识别准确率较高。
英文摘要:
      To solve the issues of time lag and operational complexity in the detection of the Mooney viscosity [ML(1+4) 100 °C] （referred to as Mooney viscosity） of natural rubber （NR） compound，a method for online prediction of the Mooney viscosity of NR compound based on a convolutional neural network （CNN） was proposed.The parameters recorded by internal mixer during the compounding process，such as the chamber temperature，mixing time，material temperature，rotor speed and torque，Mooney viscosity of plasticated rubber were organized into a matrix as input，with the Mooney viscosity of the compound as output，the CNN prediction model for the Mooney viscosity of NR compound with complex mapping relationship from input to output was constructed.The results showed that the CNN prediction model for the Mooney viscosity of NR compound was essentially convergent，with a mean square error of 0.5 between the predicted and measured values of the Mooney viscosity.The correlation coefficient（ R2） of the CNN prediction model was 0.874 3，with a recognition rate of 88.46% for prediction errors within Root number 5 ，and with a recongnition rate of 42.30% for prediction errors within Root number 2.The CNN prediction model demonstrated high indentification accuracy in predicting the Mooney viscosity of NR compounds within the range of 45～75 of Mooney viscosity of plasticated rubber.
Author NameAffiliationE-mail
LI Zhaoyang Qingdao University of Science and Technolog lishuxiadezhaoyang@163.com 
BIAN Huiguang Qingdao University of Science and Technolog  
CHEN Hailong Shandong Fengyuan Tire Manufacturing Co.，Ltd  
CHENG KUN Shandong Fengyuan Tire Manufacturing Co.，Ltd  
WANG Chuansheng Key Laboratory of Advanced Manufacturing Technology for Polymer Materials of Shandong Province；National Engineering Laboratory of Advanced Tire Equipment and Key Materials wcsmta@qust.edu.cn 
查看全文   查看/发表评论  下载PDF阅读器
关闭